When full inspection beats sampling: part 11
Sampling is the workhorse of quality control, but it is not always the right horse. Some lots are too small to sample, some defects too dangerous to estimate, and some shipments already rejected need every unit sorted. Knowing when to check every piece, and what full inspection can and cannot promise, separates professional quality planning from one-method thinking. This article maps the decision: the situations where full inspection wins, its real costs, and its honest limits.
Key takeaways
- Full inspection suits small lots, safety-critical units, sorted rework after rejection, and goods whose unit value justifies the effort.
- Acceptance sampling remains the efficient choice for stable processes and large lots, estimating lot quality from a defined sample [1].
- Documented control of the sorting process matters as much as the checking itself, since unsorted mix-ups can recontaminate verified goods [2].
- Full inspection reduces escaped defects but never reaches perfection, because inspector fatigue and method limits leave residual risk.
- The decision should be written into the quality plan with reasons, so cost and coverage stay traceable across orders.
Quality folklore says checking everything is always safer. Reality is less flattering: full inspection is slow, expensive, and still imperfect, while sampling is fast, economical, and statistically honest about its limits. Professionals choose between them deliberately, matching the method to the lot size, the defect consequence, and the process history. Amateurs default to one method for everything and pay for the mismatch in escapes or in wasted inspection days.
Sampling theory is explicit about its own boundaries. Acceptance procedures estimate lot quality from samples and suit stable production where defects are relatively rare; tiny lots, known-contaminated lots, and safety-critical units fall outside the conditions where sampling shines [1]. Documented quality systems likewise expect the control method to be planned and recorded rather than improvised per shipment [2]. Method choice is part of the plan, not an afterthought.
Part eleven gives you the map. It lists the situations favoring each method, prices the trade-offs honestly, and explains how to run full inspection so its results survive scrutiny.
Situations where every unit earns its check
Small lots top the list. When the shipment holds a few hundred units, the sample a statistical plan would draw approaches the whole lot anyway, so checking everything costs little extra and removes estimation entirely. High unit value argues the same way: jewelry components, precision instruments, and medical-adjacent goods justify per-unit attention that bulk textiles never could. Safety-critical functions form the third case, where the consequence of a single escape dwarfs inspection cost and only per-unit verification satisfies the risk logic.
Sorted rework is the fourth classic. After a lot fails sampling, the standard recovery is containing the lot and checking every unit against the failed criteria, repairing or scrapping defectives. Buyers often run this with UTS teams on site to keep the sorting honest and the records clean. The sorted lot then ships with a record of what was found, what was fixed, and what was scrapped, which beats both silent rework and blind reshipment as a basis for trust.
What full inspection honestly cannot do
Full inspection does not deliver perfection, and plans built on that belief fail twice. Inspectors fatigue: detection rates sag across long shifts of repetitive checking, and the defects that slip past tired eyes cluster exactly where attention lapsed. Methods have limits too: a visual check cannot find internal cracks, a function test cannot predict lifespan, and destructive tests cannot be applied to every unit by definition. Acknowledging these ceilings keeps expectations adult and pushes critical properties toward testing rather than toward longer staring.
Economics set the other ceiling. Per-unit checking multiplies inspector days, extends lead time, and still leaves residual risk, so applying it to large stable lots wastes money that process improvement would have spent better. The professional stance treats full inspection as a scalpel for specific situations rather than a blanket: powerful where indicated, wasteful where sampling already answers the question reliably.
Choosing the method by situation
| Situation | Better method | Reasoning |
|---|---|---|
| Large lot, stable process | Sampling | Efficient lot verdict with known statistical behavior [1] |
| Small lot under a few hundred | Full inspection | Sample would approach the lot size anyway; check everything |
| Safety-critical function | Full inspection plus testing | Single escape unacceptable; combine per-unit checks with lab evidence |
| Rejected lot needing recovery | Full sort with records | Every unit dispositioned; findings logged under document control [2] |
| High unit value goods | Full inspection | Per-unit cost trivial against escape cost |
| Stable bulk commodities | Sampling | Process history justifies estimation; spend savings on prevention |
Write the chosen method and its reasons into the order's quality plan. A method chosen silently gets second-guessed loudly after the first dispute.
Running a full sort that holds up
- Define the sort criteria preciselyName the exact defects being sorted with photo examples and limits, because sorters checking for everything effectively check for nothing.
- Separate the flows physicallySet up distinct areas and containers for unchecked, accepted, repairable, and scrap units so verified goods cannot mix back with the unsorted pile.
- Rotate sorters and verify the verifiersLimit continuous checking shifts, rotate staff, and re-check a sample of passed units to measure and control the miss rate of the sort itself.
- Record dispositions completelyLog counts of accepted, reworked, and scrapped units with reasons, keeping the record under the same document discipline as any quality file.
- Decide the lot fate in writingClose with a graded verdict on the sorted lot, stating residual limits such as properties sorting could not verify, so the buyer decides with open eyes.
After a sorted recovery, ask why sorting was needed. Full inspection rescues shipments; only process correction prevents the next rescue. Book the follow-up on the cause, not just the symptoms.
Sources
- Wikipedia: Acceptance sampling —— Overview of acceptance sampling including AQL concepts used in product inspection.
- Wikipedia: ISO 9000 —— Overview of the ISO 9000 family of quality management standards.
Frequently asked questions
Does full inspection guarantee a defect-free shipment?
No. Inspector fatigue, method limits, and untestable properties leave residual risk in every sorted lot. Full inspection sharply reduces escapes compared with shipping an unsorted rejected lot, but honest reports state what sorting covered and what it could not verify.
When is sampling actually the safer choice?
For large lots from stable processes, sampling gives a faster, cheaper verdict with explicit statistical meaning. Forcing full inspection there exhausts budgets and schedules while adding little safety. Spend the savings on process audits and prevention instead.
How do we keep sorted goods from re-mixing?
Physical separation with labeled zones and containers, plus documented handover counts at each stage. The most common sorting failure is not missed defects but verified units tumbling back into the unchecked pile. Separation discipline is the whole game.
Should the supplier pay for sorting after rejection?
Commercial terms decide, but the purchase order should state the rule in advance: who bears re-inspection and sorting cost after a failed lot. Agreed terms convert an emotional fight into an administrative step and speed up recovery.